[Nek5000-users] Stretching ratio

[nek5000-users at lists.mcs.anl.gov]

In general, the answer is no.

It does matter, however, whether you are stretching the input
grid (i.e., at some point prior to the simulation) or are stretching
the geometry (xm1,ym1,zm1) in usrdat2.

In the latter case it's rather important to stick with linear maps
of the form:

     x' = a x + b

... Otherwise you end up changing the spacing of the 
Gauss-Lobatto-Legendre points within the element and this
can lead to stability problems (e.g. a non-positive mass matrix).

When stretching the mesh (i.e., the originating SE vertices),
you of course want to resolve regions of interest, subject to the
consideration that very small mesh spacing can lead to CFL timestep
constraints if the velocity is nonzero in those regions.

Paul



On Mon, 26 Apr 2010, nek5000-users at lists.mcs.anl.gov wrote:

> Dear NEKs,
>
> when using non-uniform grids, is there any threshold/maximum for the 
> stretching ration of size of one element to an other??
>
> Best, Fred
> _______________________________________________
> Nek5000-users mailing list
> Nek5000-users at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>